Refining kernel matching pursuit

Authors:
Jianwu Li;Yao Lu
Affiliations:
Beijing Key Lab of Intelligent Information Technology, School of Computer, Beijing Institute of Technology, Beijing, China;Beijing Key Lab of Intelligent Information Technology, School of Computer, Beijing Institute of Technology, Beijing, China
Venue:
ISNN'10 Proceedings of the 7th international conference on Advances in Neural Networks - Volume Part II
Year:
2010

Citing 5
Cited 0

Kernel Matching Pursuit

Machine Learning
Kernel matching pursuit classifier ensemble

Pattern Recognition
Kernel matching pursuit for large datasets

Pattern Recognition
LIBSVM: A library for support vector machines

ACM Transactions on Intelligent Systems and Technology (TIST)
Base vector selection for kernel matching pursuit

ADMA'06 Proceedings of the Second international conference on Advanced Data Mining and Applications

Quantified Score

Hi-index	0.00

Visualization

Abstract

Kernel matching pursuit (KMP), as a greedy machine learning algorithm, appends iteratively functions from a kernel-based dictionary to its solution An obvious problem is that all kernel functions in dictionary will keep unchanged during the whole process of appending It is difficult, however, to determine the optimal dictionary of kernel functions ahead of training, without enough prior knowledge This paper proposes to further refine the results obtained by KMP, through adjusting all parameters simultaneously in the solutions Three optimization methods including gradient descent (GD), simulated annealing (SA), and particle swarm optimization (PSO), are used to perform the refining procedure Their performances are also analyzed and evaluated, according to experimental results based on UCI benchmark datasets.